Among databases (DB), there is a DB (for example, schema-full DB) that operates under conditions having schemas in which an item name and a data type have been defined in advance for each item of the DB. Moreover, there is a DB (for example, schemaless DB) that operates under conditions having no schemas. Furthermore, there is a DB (for example, hybrid DB) that operates under conditions in which a schema-full DB and a schemaless DB both exit. Specifically, in the hybrid DB, while common attributes that are given to items that structure a table are defined in a schema, attributes of some items are defined ad hoc in a schema.
For the hybrid DB, a technique of extending a schema when multiple database sources are merged has been disclosed (for example, Japanese Laid-open Patent Publication No. 2006-107466). For example, when a predetermined application merges multiple data sources, fields in each data source are mapped. Specifically, the first data source is extracted as the initial data source, fields of each data source subsequent to the extracted data source are associated with fields of the initial data source, and fields in the subsequent data sources that are not included in the initial data source are added. The fields mapped from each input data source are added ad hoc in the schema.
However, in the technique of extending a schema, there is a problem that when an unintended item is added to a schema, the unintended item is not identified. That is, an item that is added to the schema by schema extension is added without checking the item name, the data type, and the like. Accordingly, an unintended item can be added, for example, by making a spelling mistake in an item name. Therefore, in a conventional predetermined application, it is impossible to determine whether an added item is an intended item or an unintended item.